## round off problem in matlab

### Ali Asghar (view profile)

on 19 Sep 2019
Latest activity Commented on by Ali Asghar

on 23 Sep 2019

### Steven Lord (view profile)

I have 4 variables each have 4 values (1x4) in it and I want to store all in value in 1 variable (4x4) but values are change when i do it.
MAV = [0.0924317411754942 0.0313003411430704 0.00365072354007141 0.00332056432198707]
waveLen = [172.697769103870 68.7603283596702 69.2536410243476 63.2141284111186]
zcdOut = [560 719 6155 6260]
slope_count = [478 212 4949 4830]
HG0_extracted_feature = [waveLen; MAV; zcdOut; slope_count]
it gives
173 69 69 63
0 0 0 0
560 719 6155 6260
478 212 4949 4830
note MAV values become zero, it round off waveLen, etc values...
how i correct it?
Thank you

### Steven Lord (view profile)

on 19 Sep 2019

Based on that display I suspect one or both of the variables zcdOut or slope_count are stored in an integer class. When you combine data stored in an integer class with data stored in a non-integer class, "all elements of the resulting matrix are given the data type of the left-most integer".
Consider converting the variable stored in an integer class to double before combining them into one array, so that the resulting matrix is also a double.

Show 1 older comment
Stephen Cobeldick

on 22 Sep 2019
Steven Lord

### Steven Lord (view profile)

on 22 Sep 2019
Check the types of variables using class or whos.
As Stephen Cobeldick said, to convert variables to double use the double function.
Ali Asghar

on 23 Sep 2019
thanks dear

### Raj (view profile)

on 19 Sep 2019

format longg
MAV = [0.0924317411754942 0.0313003411430704 0.00365072354007141 0.00332056432198707]
waveLen = [172.697769103870 68.7603283596702 69.2536410243476 63.2141284111186]
zcdOut = [560 719 6155 6260]
slope_count = [478 212 4949 4830]
HG0_extracted_feature = [waveLen; MAV; zcdOut; slope_count]

Ali Asghar

### Ali Asghar (view profile)

on 20 Sep 2019
Dear i have data called filter_datawindow 24000x4.
i apply functions in loop that store the value of waveLen, MAV, zcdOut, slope_count in workspace in long format but when i merge all values is give same result as
173 69 69 63
0 0 0 0
560 719 6155 6260
478 212 4949 4830
for i = 1:4
waveLen(i) = find_waveform_length(filter_datawindow(:,i))
MAV (i)=find_MAV(filter_datawindow(:,i),2.4e4)
zcdOut (i) = zcd(filter_datawindow(:,i));
slope_count (i) = find_slopSign(filter_datawindow(:,i),0.001)
end
format longg
HG0_extracted_feature = [waveLen; MAV; zcdOut; slope_count]
gives
173 69 69 63
0 0 0 0
560 719 6155 6260
478 212 4949 4830

on 19 Sep 2019
Edited by Matt J

### Matt J (view profile)

on 19 Sep 2019

The values are not changed. They are just displayed to the screen to 4 decimal points. Use format long to see more precision. Also, note that all elements are premultiplied by 1000.

Ali Asghar

### Ali Asghar (view profile)

on 20 Sep 2019
when i merge all values it gives
173 69 69 63
0 0 0 0
560 719 6155 6260
478 212 4949 4830
172.69776910387 68.7603283596702 69.2536410243476 63.2141284111186
0.0924317411754942 0.0313003411430704 0.00365072354007141 0.00332056432198707
560 719 6155 6260
478 212 4949 4830
Matt J

### Matt J (view profile)

on 20 Sep 2019
As Steven. mentioned, that is probably because one of the vectors you are merging is uint8 or similar.
Steven Lord

### Steven Lord (view profile)

on 20 Sep 2019
Rows 3 and 4 of the resulting array have values too large for uint8 or int8 so they would need to be at least 16-bit integers.